29/06/2022

Brisbane Flooding Hackathon Presentation

For R Ladies Brisbane


Motivation

Impacts beyond flood inundation

Access to services - shops, hospitals, petrol stations

Loss of power

Road and public transport closures


What is the gap?

What is vulnerability?

Vulnerability = Exposure + Sensitivity + Adaptive Capacity
IPCC Climate Change 2001: Impacts, Adaptation, and Vulnerability (eds McCarthy, J. J., Canziani, O. F., Leary, N. A., Dokken, D. J. & White, K. S.)

Exposure is the immediate effect

Sensitivity represents the longstanding innate properties of the population

Adaptive capacity are the things that can be done to change impacts

Our approach

We aim to create a vulnerability index for the Brisbane City local government area (LGA) predicting the vulnerability of each level 2 statistical area (SA2).

Exposure: flood extent intersection with properties, services, infrastructure

  • rules for certain services like power, buffer of affected area
  • various levels of exposure
  • measure changed access to services with flooded roads

Sensitivity: SA2 averages of demographic and socio-economic census data

Adaptive capacity: actions in place to increase resilience to flooding

  • buy back programs
  • moving towns, Grantham
  • minimum build heights

R Libraries

# simple features package for vector
# spatial data
library(sf)
# for data manipulation
library(tidyverse)
# Australian maps
library(ozmaps)

Code is courtesy of Rich Cotrell

Great data from Queensland Spatial Catalogue - QSpatial

Flood awareness plot

flood_extents_shp <- st_read(paste0(path,
  "spatial/Flood__Awareness__Extents/Flood___Awareness___Extents.shp"))

ggplot() + geom_sf(data = flood_extents_shp,
  fill = "lightblue", colour = "grey90",
  size = 0.01) + theme_bw()

Local Government Areas - {ozmaps}

lgas <- ozmaps::ozmap_data(data = "abs_lga")

bris_lga <- lgas %>%
  filter(NAME == unique(lgas$NAME)[grep("Brisbane",
    unique(lgas$NAME))]) %>%
  st_transform(crs = st_crs(flood_extents_shp))

Level 2 Statistical Area

From the Australian Bureau of Statistics:

SA2_aus <- st_read(paste0(path, "spatial/sa2_2016_aust_shape/SA2_2016_AUST.shp"))

SA2_bris <- SA2_aus %>%
  filter(STE_NAME16 == "Queensland" &
    GCC_NAME16 == "Greater Brisbane" &
    SA4_NAME16 %in% c("Brisbane - East",
      "Brisbane - North", "Brisbane - South",
      "Brisbane - West", "Brisbane Inner City") &
    (!SA2_NAME16 %in% c("Redland Islands"))) %>%
  st_transform(crs = st_crs(flood_extents_shp))

Plot LGAs and SA2s

ggplot() + geom_sf(data = bris_lga) + labs(title = "LGA") +
  theme(plot.title = element_text(size = 30)) +
  theme_bw()
ggplot() + geom_sf(data = SA2_bris) + labs(title = "SA2s") +
  theme(plot.title = element_text(size = 30)) +
  theme_bw()

Emergency Services

ems <- st_read(paste0(path, "/spatial/QSC_QLD emergency service locations/Emergency_services_facilities.shp")) %>%
  st_transform(crs = st_crs(flood_extents_shp))

ems_crop <- st_crop(x = ems, y = st_bbox(flood_extents_shp)) %>%
  filter(ATTRIBUTES %in% c("Queensland Ambulance Service",
    "Queensland Fire and Emergency Services",
    "Queensland Police Service"))

ems_intersect <- st_intersection(ems_crop,
  SA2_bris)

Plot Emergency Services

ggplot() + geom_sf(data = SA2_bris, alpha = 0,
  size = 0.5) + geom_sf(data = ems_intersect,
  aes(colour = ATTRIBUTES), size = 1) +
  theme_bw()

Petrol stations

petrol <- st_read(paste0(path, "spatial/Petrol_Stations_Australia/PetrolStations.gdb")) %>%
  st_transform(crs = st_crs(flood_extents_shp))

petrol_crop <- st_crop(x = petrol, y = st_bbox(SA2_bris))
petrol_intersect <- st_intersection(x = petrol_crop,
  y = SA2_bris)

Putting it all together

Flood extent with emergency services and petrol stations

Work in Progress

How to turn this in to an index?

  • Combine all the variables of each component (Exposure etc) into a score
  • Add exposure, sensitivity, adaptive capacity and scale

Future Work - Dream RShiny App!

Example: SpatialEpiApp for disease surveillance by Paula Moraga.


SpatialEpiApp is a Shiny web application that allows to visualize spatial and spatio-temporal disease data, estimate disease risk, and detect clusters.

Team

Catherine Kim Kaitlyn Brown Richard Cottrell Jess Hopf Aiden price Connie Susilawati

Thank you!